基于小波包分析和高阶模糊神经网络的滚动轴承故障诊断

滚动轴承是旋转机械中最易发生故障的元件之一,提出了一种基于小波包分析和高阶模糊BP神经网络的滚动轴承故障诊断新方法。该方法的具体诊断过程:采用小波包分解的方法提取样本信号各频段的Shannon熵值并结合其他一些量化指标,经筛选后作为特征向量输入滚动轴承故障诊断高阶模糊神经网络,对该网络进行训练与检验。实验表明这种方法与传统方法相比,在收敛速度及对训练总误差控制方面具有更大的优越性。

Fault Diagnosis of Rolling Bearing Based on Wavelet Packet Analysis and High-order Fuzzy High-order Fuzzy Neural Network

The rolling bearing is one element which most easily to go wrong of rotating machinery.A new method of rolling bearing fault diagnosis based on wavelet packet analysis and high-order fuzzy BP neural network is proposed.Diagnosis concrete process of the method is as follows: wavelet packet decomposition is adopted to extract Shannon entropy of sample signals from each frequency band,and which combining with other quantitative index served as eigenvector to input rolling bearing fault diagnosis based on high-order fuzzy neural network after be selected.Then training and testing is carried out.The experimental results show the method has advantage over the routine method in convergence rate and training total error controlling.